FAQ - The Do's and Don'ts in Mistral-Le Chat
The use of AI must comply with transparency, diligence, non-discrimination, and data protection. Whether AI can be used depends on the purpose of use and the input data.
FAQ Education - FAQ General - FAQ Research
These FAQs serve as a guideline for frequently asked questions and have been prepared in consultation with AJZ. Some nuances have been omitted for readability. If you are unsure whether something is allowed, please contact your P&S coordinator for more information. Note:
Attention: The FAQs have been compiled for the Mistral-Le Chat Enterprise licence of the University of Groningen. This licence is only active on the @rug.nl account. Accounts with other email addresses are not covered by the UG's Enterprise license. You can recognize the Enterprise licence by checking whether it is mentioned in the top left corner after logging in. Other External AI tools may only be used for processing public information.
What type of personal data (degree of confidentiality) may I process in Le Chat?
|
Confidentiality
|
High
|
Medium
|
Low
|
Public
|
|
Special personal data*
Intellectual property (IP)** |
Personal data
Organization sensitive information |
Organizational information |
Public infor- |
|
|
Mistral -
Le Chat |
No
|
Be cautious
|
Yes
|
Yes
|
|
Other External AI tools: ChatGPT, Claude, Perplexity etc,
|
No
|
No
|
No
|
Yes
|
What data is permitted?
|
Data Type
|
Examples
|
Allowed?
|
|
No personal data
|
Anonymized or personal data removed.
|
Yes (including all tools)
|
|
Ordinary personal data or confidential business information
|
Names, email addresses, general education information, course codes, board meetings, reorganization plans.
|
Yes, provided the code interpreter, image generation, web search, and Flash Answers are disabled.
|
|
Special categories of personal data
|
Sensitive personal data: race/ethnicity, political opinions, religion/philosophical beliefs, trade union membership, genetic data, biometric data, health data, sexual behaviour/orientation.
|
No
|
|
Sensitive personal data
|
Student results, assessments, and grades.
|
No
|
|
Intellectual property (IP)
|
Information regarding inventions.
|
No
|
What type of data (degree of confidentiality) may I process in Le Chat?
Overview:
|
Confidentiality
|
High
|
Medium
|
Low
|
Public
|
|
Special personal data*
Intellectual property (IP)** |
Personal data
Organization sensitive information |
Organizational information |
Public infor- |
|
|
Mistral -
Le Chat |
No
|
Be cautious
|
Yes
|
Yes
|
|
Other External AI tools: ChatGPT, Claude, Perplexity etc,
|
No
|
No
|
No
|
Yes
|
*Special personal data includes information about race/ethnicity, political views, religion/philosophy, union membership, genetic data, biometric data (for unique identification), health data, and/or sexual behaviour/orientation.
**Sensitive information about intellectual property, such as confidential patent applications, unpublished innovations, and scientific articles, especially during the development phase or before formal protection (e.g., patent, trademark registration) is obtained.
Mistral - Le Chat: Be cautious at medium confidentiality level: Mistral can be used, provided that code interpreter, image generation, web search, and Flash Answers are disabled. Click on "Tools" and disable the functions (see example below).
ChatGPT, Claude, Perplexity, etc.: These tools are not supported by the UG and have not all been thoroughly assessed for privacy and security risks. In some cases, these tools have even been rejected due to the risks they pose. External AI tools may only be used for processing public information.

What data is permitted?
What data is permitted?
|
Data Type
|
Examples
|
Allowed?
|
|
No personal data
|
Anonymized or personal data removed.
|
Yes (including all tools)
|
|
Ordinary personal data or confidential business information
|
Names, email addresses, general education information, course codes, board meetings, reorganization plans.
|
Yes, provided the code interpreter, image generation, web search, and Flash Answers are disabled.
|
|
Special categories of personal data
|
Sensitive personal data: race/ethnicity, political opinions, religion/philosophical beliefs, trade union membership, genetic data, biometric data, health data, sexual behaviour/orientation.
|
No
|
|
Sensitive personal data
|
Student results, assessments, and grades.
|
No
|
|
Intellectual property (IP)
|
Information regarding inventions.
|
No
|
FAQ - Education
Can I use Mistral-Le Chat to provide feedback on assignments such as essays and theses from students?
Can I use Mistral-Le Chat to provide feedback on assignments such as essays and theses from students?
Yes, as long as no special personal data or intellectual property is processed in the theses, and the AI feedback does not play an automatic and unclear role in decision-making/evaluation and only serves to support the teacher. The teacher remains responsible for assessments, and the use of AI must be communicated to students.
Can I use Mistral-Le Chat to develop educational materials?
Can I use Mistral-Le Chat to develop educational materials?
Yes, as long as the materials do not contain intellectual property or special categories of personal data.
Can I process personal data of students using Mistral-Le Chat?
Can I process personal data of students using Mistral-Le Chat?
Yes, as long as the data does not contain special personal data. Ensure that there is no sensitive information such as grades, study results, and evaluations, or special personal data such as information about race/ethnicity, political views, religion/philosophy, union membership, genetic data, biometric data (for unique identification), health data, and/or sexual behaviour/orientation in the notes.
Analyzing study performance is not allowed in Mistral Le Chat, as this processing poses too high risks. For processing personal data of students and employees, Mistral - Le Chat is suitable, provided that the use of code interpreter, image generation, web search, and Flash Answers is disabled.
Overview - What personal data is permitted?
Can I analyze student study results with Mistral-Le Chat?
Can I analyze student study results with Mistral-Le Chat?
No, analyzing student study results is not allowed because this information (in combination with other information about students) may need to be considered as special personal data.
Do I need to explicitly ask my students for permission if I intend to use a Mistral-Le Chat agent to provide feedback on their written work?
Do I need to explicitly ask my students for permission if I intend to use a Mistral agent to provide feedback on their written work?
Theses, exams, and other student work are almost always considered personal data, even without a name or student number, due to traceability via context or writing style. Processing can therefore only take place on the basis of a legal ground. Consent is not suitable here; the UG can rely on its educational and examination task (Article 6(1)(e) GDPR). Students must, however, be informed, for example via the syllabus and privacy statement.
The AI Act stipulates that decisions affecting academic progress must never be made solely by AI: there must always be a ‘human in the loop’. AI may support, summarize, and suggest improvements, but it may not assess or decide.
Additionally, copyright remains with the student. Their work may only be used for educational purposes, not for training or external publication.
Recommended communication to students:
“The work you submit (thesis, exam, assignment) may be analysed by lecturers using Mistral to generate feedback. This is done within the processing agreement that UG has concluded with Mistral. Copyright remains fully yours. If you disagree with the feedback, you can contact your lecturer for a possible second review.”
Does the Mistral-Le Chat pilot also provide access for students, so that working with an AI assistant can be integrated into education?
Does the Mistral-Le Chat pilot also provide access for students, so that working with an AI assistant can be integrated into education?
At present, access to Mistral-Le Chat for students within UG’s AI pilot is not yet available. CIT is coordinating this with UG’s management, focusing particularly on the additional burden on the AI Office due to support requests and the risk of vendor lock-in when integrating AI into education.
Once these points have been considered and clear agreements have been made, access for students can be opened. The goal is to deploy AI in education in a responsible and practical manner.
May I upload a PDF of a published scientific article to a Library in Mistral–Le Chat?
May I upload a PDF of a published scientific article to a Library in Mistral–Le Chat?
At present, there is no definitive answer regarding whether non-open access articles may be uploaded to Mistral–Le Chat, even if the University of Groningen (UG) holds a subscription to the journal. The parties involved (University Library, CIT, and AJZ) are working on a practical and legally sound solution.
This is a complex issue that requires greater knowledge and awareness of licences and the various types of licences applicable to scientific publications. It is important for UG staff and researchers to be aware of the following:
-
The rights and restrictions associated with different licence types (e.g., closed subscriptions, open access, hybrid models).
-
How licence agreements with publishers work and what is and is not permitted.
-
What alternatives exist for sharing and analysing articles within the applicable licence terms.
FAQ - General use
How can I create an account or log in?
How can I create an account or log in?
You can create an account with Mistral-Le Chat by logging in with your UG email address. (Use your @rug.nl email address, click ‘Continue’, and then select ‘Continue with University of Groningen – SSO’.) It is important not to use Google, Apple, or Microsoft to log in. Use your P-number and corresponding password, and enter the one-time code.
Can I process transcripts using Mistral-Le Chat?
Can I process transcripts using Mistral-Le Chat?"
Processing transcripts in Mistral-Le Chat is allowed, provided the following conditions are met:
-
The transcripts do not contain special personal data.
-
The tools 'code interpreter', 'image generation', and 'web search' are disabled.
-
The 'Flash Answers' function is disabled.
Transcripts that contain only ordinary personal data, such as names, general education information, and course codes, may be processed. However, if the transcripts contain sensitive information such as study results or special personal data such as information about race/ethnicity, political views, religion/philosophy, union membership, genetic data, biometric data (for unique identification), health data, and/or sexual behaviour/orientation, it is not allowed.
Can I have audio files transcribed by Mistral-Le Chat?
Can I have audio files transcribed by Mistral-Le Chat?”
Transcribing audio files will be launched soon (October) by Mistral-Le Chat. Processing recordings in this programme is allowed, provided the following conditions are met:
-
The transcripts do not contain special personal data.
-
The tools 'code interpreter', 'image generation', and 'web search' are disabled.
-
The 'Flash Answers' function is disabled.
Recordings that contain only ordinary personal data, such as names, general education information, and course codes, may be processed. However, if the recordings contain sensitive information such as study results or special personal data such as information about race/ethnicity, political views, religion/philosophy, union membership, genetic data, biometric data (for unique identification), health data, and/or sexual behaviour/orientation, it is not allowed.
Can I convert meeting notes into minutes using Mistral-Le Chat?
Can I convert meeting notes into minutes using Mistral-Le Chat?
Yes, that is allowed, as long as the notes do not contain any special categories of personal data. You may process ordinary personal data, such as names and general discussion points. Ensure that no sensitive information—such as grades, academic results, assessments, or special categories of personal data (e.g., information about race/ethnicity, political opinions, religion/philosophical beliefs, trade union membership, genetic data, biometric data for unique identification, health data, and/or sexual behaviour/orientation)—is included in the notes.
For meetings involving organization-sensitive information, Mistral-Le Chat is suitable, provided that the use of the code interpreter, image generation, web search, and Flash Answers is disabled (see the example under Mistral-Le Chat: Caution at medium confidentiality level).
Can I summarize or textually or substantively improve university policy documents using Mistral-Le Chat?
Can I summarize or textually or substantively improve university policy documents using Mistral-Le Chat?
Yes, you can.
Can I use Mistral-Le Chat to prepare presentations?
Can I use Mistral-Le Chat to prepare presentations?
Yes, as long as no special personal data or intellectual property is processed in the presentations.
Can I use Mistral-Le Chat to write emails on behalf of the university?
Can I use Mistral-Le Chat to write emails on behalf of the university?
Yes, as long as the content of the emails does not contain special personal data or intellectual property.
Can I use Mistral-Le Chat to plan and organize meetings?
Can I use Mistral-Le Chat to plan and organize meetings?
Yes, as long as no special personal data of participants is processed.
Can I use Mistral-Le Chat on my tablet or mobile phone?
Can I use Mistral-Le Chat on my tablet or mobile phone?
The mobile app currently does not work for enterprise users (such as the UG) because the Mistral app does not support the Single Sign-On (SSO) login method, which is required to comply with UG’s privacy and security guidelines. Mistral is developing this functionality and expects to roll it out soon.
Does the pilot also provide access to Mistral-Le Chat’s models via an API?
Does the pilot also provide access to Mistral’s models via an API?
The current Mistral – Le Chat pilot at the UG does not yet offer direct access to Mistral’s models via an API. This functionality depends on further coordination between CIT, the faculties, and other involved services.
Efforts are actively underway to realize this possibility in the future. However, due to the complexity and numerous dependencies, it is currently impossible to provide a concrete timeline for when, how, and for whom this functionality will become available.
FAQ - Research
Can I check research papers for plagiarism with Mistral - Le Chat?
Can I check research papers for plagiarism with Mistral - Le Chat?
No, checking for plagiarism is outside the scope of Mistral - Le Chat, and specialized tools are available for this.
See EDU-support for more information on Assignment with plagiarism detection.
Can I check research papers for AI use with Mistral - Le Chat?
Can I check research papers for AI use with Mistral - Le Chat?
No, checking for AI use is outside the scope of Mistral - Le Chat. Due to, among other things, the high risk of false positives and false negatives, scores from AI detection tools do not count as evidence of student fraud, according to the UG AI policy for education.
See EDU-support for more information on "Dealing with fraud and risks.
May I upload a PDF of a published scientific article to a Library in Mistral–Le Chat?
May I upload a PDF of a published scientific article to a Library in Mistral–Le Chat?
At present, there is no definitive answer regarding whether non-open access articles may be uploaded to Mistral–Le Chat, even if the University of Groningen (UG) holds a subscription to the journal. The parties involved (University Library, CIT, and AJZ) are working on a practical and legally sound solution.
This is a complex issue that requires greater knowledge and awareness of licences and the various types of licences applicable to scientific publications. It is important for UG staff and researchers to be aware of the following:
-
The rights and restrictions associated with different licence types (e.g., closed subscriptions, open access, hybrid models).
-
How licence agreements with publishers work and what is and is not permitted.
-
What alternatives exist for sharing and analysing articles within the applicable licence terms.